Estimation of Component Reliability in Coherent Systems With Masked Data

Abstract

The reliability of a coherent system of components depends on the reliability of each component and the initial statistical work should be an estimation of the reliability of each component. This paper represents a challenging task because if the system fails, the failure time of a given component cannot be observed, that is, the phenomenon of censored data occurs. A solution for the reliability estimation of components exists when the system failure time and the status of each component are available at the time of system failure. However, it may be difficult to identify the status of the components at the moment of system failure. Such cases represent systems with masked causes of failure. Since parallel and series systems are the simplest systems, numerous solutions have been reported in the literature. To the best of our knowledge, this paper is the first to present the general case of coherent systems without the restriction of an identically distributed lifetime. The three-parameter Weibull Bayesian model is proposed. The Gibbs with the Metropolis-Hasting algorithm supports the statistical work of obtaining the posterior distribution quantities. With several simulations, the excellent performance of the model is evaluated. A real dataset of computer hard drives is analyzed to show the practical relevance of the proposed model.